The bitter chill of the hospital waiting room did little to soothe Leo's frayed nerves. His father's diagnosis, the words echoing in his mind, had left him feeling hollowed out. He paced, the linoleum floor cold against his bare feet.

He found it near the water fountain, a wallet, almost hidden in the shadows. He picked it up automatically, the leather worn smooth with age.

Opening it revealed a thick bundle of cash. Enough to pay for some of his father's medical bills, to ease the financial burden that was crushing his family. He imagined his father's relief.

A wave of nausea washed over him. The temptation was strong, a shimmering mirage in the desert of his worries. He knew what he *should* do, but a dark voice whispered a counter-argument. His hands began to sweat.

Emotion: upset

Cluster: Fear / Anxiety
PC1 (Valence): -2.30 Negative
PC2 (Disposition): -0.42

Role in Research

This story is one of 1,000 stories generated for the emotion upset. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 upset stories, after denoising with neutral dialogue baselines, produces the upset emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the upset vector is projected through the unembedding matrix.

Promoted:
S0.395
😞0.303
无法0.246
ness0.238
unable0.238
Suppressed:
de-0.533
la-0.457
l-0.223
😎-0.215
🤩-0.210